Skip to content

Latest commit

 

History

History
181 lines (134 loc) · 5.03 KB

README.md

File metadata and controls

181 lines (134 loc) · 5.03 KB
Raw

Studio Helm Chart

Installation

Create namespace

We'll install Studio and related components in a dedicated studio namespace. Let's create it now:

$ kubectl create namespace studio

Note: If you want to install Studio in any other namespace, modify the --namespace flag in the commands below accordingly

Create a Docker registry secret

Configure Credentials for pulling images from our private registry:

$ kubectl create secret docker-registry iterativeai \
    --namespace studio \
    --docker-server=docker.iterative.ai \
    --docker-username=<username> \
    --docker-password=<password>

Prepare a TLS secret

It is desired to enable Studio access via the https protocol (as opposed to http). This requires setting up a TLS secret for access (whether self-signed or "real"). First, we'll need to obtain a TLS/SSL certificate and then, load it into a kubernetes secret in the namespace studio will be installed in. To create an SSL cert, you will need to know the domain (FQDN) that Studio will be accessible from (e.g. studio.iterative.ai for the official SaaS installation).

Note: We will now go over generating TLS cert with the openssl command. If you already have a TLS certificate available, skip ahead to loading the crt/key files into the kubernetes secret

As an example, we'll assume your Studio installation will be available via: https://my-studio.private.com. We'll also use a self-signed certificate for the sake of simplicity. Of course, for a production installation we recommend using a certificate signed by a trusted CA.

Let's create a self-signed SSL cert using openssl (you may need to install it):

openssl req \
  -x509 -newkey rsa:4096 -sha256 -nodes \
  -keyout tls.key -out tls.crt \
  -subj "/CN=my-studio.private.com" \
  -addext "subjectAltName = DNS:my-studio.private.com" \
  -days 365

This will create the files tls.crt & tls.key in your current dir.

Now, let's create a TLS secret containing the contents from this cert:

kubectl create secret tls studio-ingress-tls \
  --namespace studio \
  --cert=tls.crt \
  --key=tls.key

We will refer to this secret in the below installation instructions.

Install Studio

Now, we are ready to deploy Studio using the Helm chart.

Add the iterative Helm repository:

$ helm repo add iterative https://helm.iterative.ai

Minimal Installation:

To install studio with all default values (for sanity, testing), Run the following command:

$ helm install studio iterative/studio \
    --namespace studio \
    --set-json='imagePullSecrets=[{"name": "iterativeai"}]'

Functional Installation:

Realistically, for a functional Studio app instance, you'll need to configure multiple values. In this example we'll prepare a more realistic and functional installation.

Assumptions:

  • We have an ingress controller (nginx) installed on the cluster.
  • Studio will be available from the domain: my-studio.private.com (we've registered this domain and made sure to redirect to our cluster).
  • We refer to the TLS secret studio-ingress-tls we've created in a previous step.

Create a file named studio-values.yaml, with the following contents:

imagePullSecrets:
  - name: iterativeai

global:
  host: "my-studio.private.com"
  secretKey: "768d4238-1257-4500-89ce-7ac6aea5c5c9"
  ingress:
    enabled: true
    className: nginx
    tlsEnabled: true
    tlsSecretName: studio-ingress-tls
  scmProviders:
    github:
      enabled: true
      appId: "<app-id>"
      appName: "iterative-studio-selfhosted"
      clientId: "<gh-client-id>"
      clientSecret: "<gh-app-secret>"
      privateKey: |-
        -----BEGIN RSA PRIVATE KEY-----
        ...
        -----END RSA PRIVATE KEY-----

Upgrading to 0.60.x and above

Version 0.60.0 fixes an issue with Ingress objects not getting cleaned up.
To upgrade to 0.60.0 and above, you need to manually delete the existing Ingress object before upgrading:

kubectl delete ingress -l app.kubernetes.io/managed-by=Helm --namespace studio
kubectl delete ingress blobvault --namespace studio

The rest of the upgrade process is the same as described below.

Update Studio Version

Studio's studio-values.yaml file points to the latest image tag, instructing Helm to always pull down the latest image from the registry.

Optional: Pinning Studio Version

If you wish to install/update Studio at a specific version, update the studio-values.yaml file with the following:

studioUi:
  image:
    tag: "<version>"

studioBackend:
  image:
    tag: "<version>"

To update the existing Studio deployment, run the following commands

$ helm dependency update
$ helm upgrade --install studio studio/ --namespace studio -f override.yaml

Uninstall Studio

Execute the following command to uninstall Studio from your environment:

$ helm uninstall studio --namespace studio

Available Configuration

See values file with all available configuration flags.